study of homogeneous mineral acid dehydration of
TRANSCRIPT
Syracuse University Syracuse University
SURFACE SURFACE
Dissertations - ALL SURFACE
December 2015
Study of homogeneous mineral acid dehydration of Study of homogeneous mineral acid dehydration of
monosaccharides in a CSTR monosaccharides in a CSTR
Siddharth Sharad Bhat Syracuse University
Follow this and additional works at: https://surface.syr.edu/etd
Part of the Engineering Commons
Recommended Citation Recommended Citation Bhat, Siddharth Sharad, "Study of homogeneous mineral acid dehydration of monosaccharides in a CSTR" (2015). Dissertations - ALL. 360. https://surface.syr.edu/etd/360
This Thesis is brought to you for free and open access by the SURFACE at SURFACE. It has been accepted for inclusion in Dissertations - ALL by an authorized administrator of SURFACE. For more information, please contact [email protected].
Abstract
Simple sugars (Aldohexoses/pentoses), (Ketohexoses/pentose) when subjected to a dehydration
reaction, produce various compounds. An example of one such compound is 5-hydroxymethyl
furfural. Such chemical products may be used in subsequent processing steps (hydrolysis, aldol
condensation, hydrogenation and dehydration) to produce similarly structured condensed
compounds that form the basis of complex molecules used in various industries such as
production of fuels, chemical reagents and fertilizers. Most studies focus on heterogeneous
packed beds and batch reactors to carry out dehydration reactions. This study illustrates use of a
Continuous Stirred-Tank reactor to carry out such dehydration reactions. In order to maintain the
homogenous nature of the reaction, the catalyst chosen was sulfuric acid. Use of a CSTR would
allow study of kinetics and yields for varying residence times. This data could be used to design
large scale operations in biomass processing. This study is aimed at investigating variables such
as changing physical properties of the fluid reactant during reaction, variation in pH and
consequent change in proton concentrations. By understanding the impact of these variables,
more accurate rate measurements can be made. In the process of developing the right method for
data acquisition, several substantial changes were made to the reactor and methods. These
changes were instrumental in development of a cyclic process of data collection and changing
methods and design, thus highlighting the chemical engineering heuristics approach and
refinement in processes. The sequence of corrected results give us better insight into the process
and help develop accurate models for the reactions in the scope of this thesis and also future
studies. The presented results of rates and yields may be used to develop processes based on
requirement and feasibility.
STUDY OF HOMOGENEOUS MINERAL ACID DEHYDRATION OF
MONOSACCHARIDES IN A CSTR
By
Siddharth Bhat
Mumbai University 2009-2013
THESIS
Submitted in partial fulfillment of the requirements for the
Degree of Master of Science in Chemical Engineering
In the Graduate School of Syracuse University
December 2015
Copyright © Siddharth Bhat
All Rights Reserved
iv
Acknowledgements
I would like to thank my advisor, Professor Jesse Q. Bond for granting me the
opportunity to learn and grow as a part of the lab. My journey through the Masters program has
been especially memorable and enjoyable on account of the continued guidance and support I
received. I am grateful to all the faculty members, administrative staff and workshop staff (Bill
and Dick) that assisted me through these two years. I appreciate the time spent by Dr. Benjamin
Akih Kumgeh, Dr. Shikha Nangia and Dr. Lawrence Tavlarides for taking the time to review my
written thesis and for being a part my dissertation panel.
A special thank you is directed to my lab mates, especially Omar who designed the
reactor setup I had the privilege of working on, for the time and effort he spent in guiding me
through the basics of laboratory operations, Argy for his valuable advice. Thanks to Christian,
Josh and Shoufau for always keeping it simple and fun.
Lastly, I would like to thank my entire family for their unconditional love and support
through my education. I would like to dedicate the success of this thesis project to my parents,
Sharad Bhat and Veena Bhat and to my girlfriend Niyati Shetty.
v
Table of Contents
Acknowledgements……………………………………………………………………………….iv
List of illustrative materials………………………………………………………………………vi
Figures………………………………………………………………………………...vi
Tables………………………………………………………………………………...viii
Chapter I Introduction…………………………………………………………………………..1
Chapter II Literature Review…………………………………………………………………...3
2.1 Biomass……………………………………………………………………………3
2.2 Upgrading Technologies…………………………………………………………12
2.3 Acid Catalyzed Dehydration……………………………………………………..14
2.4 Platform Chemicals………………………………………………………………17
2.4.1 HMF…………………………………………………………………...........17
2.4.2 Furfural……………………………………………………………………...19
2.5 Reaction Orders and Rate Laws………………………………………………….21
Chapter III Layout of CSTR and Design……………………………………………………..23
Chapter IV Mixing……………………………………………………………………………...25
4.1 Experimental Methods (Mixing)…………………………..……………………..25
4.2 Mixing Results and Discussion……...…………………………………………...29
Chapter V Dehydration Experiments…………………………………………………………32
5.1 Analytical Equipment and Chemicals……………………………………………32
5.2 Selection of reactor………………………………………………………………32
5.3 Experimental Procedure………………………………………………………….34
5.4 Analytical Methods………………………………………………………………35
5.5 Kinetic Results and Discussion…...……………………………………………...38
5.5.1 Glucose……………………………………………………………………...38
5.5.2 Fructose……………………………………………………………………..48
5.5.3 Xylose……………………………………………………………………….52
Chapter VI Conclusion and Perspectives …………………………………………………….57
References……………………………………………………………………………………….58
Vita………………………………………………………………………………………………62
vi
A) Figures
Figure 1: Composition of Biomass……………………………………………………………….3
Figure 2: Structure of Cellulose…………………………………………………………………..5
Figure 3: Components of Hemicellulose…………………………………………………………5
Figure 4: Components of Starch (Amylose and Amylopectin) ………………………………….6
Figure 5: Structure of Furanic Aldehydes (Furfural and HMF) …………………………………7
Figure 6: Structure of monolignol units…………………………………………………………..8
Figure 7: Biomass Processing/Upgrading routes…………………………………………………9
Figure 8: Sugar monomers Glucose, Fructose and Xylose……………………………………...10
Figure 9: Fischer projections of open chain and closed chain forms of glucose………………..11
Figure 10: Postulated mechanism for the production of hydroxymethylfurfural (HMF) 68….....15
Figure 11: HMF as a platform chemical………………………………………………………...17
Figure 12: Furfural as a platform chemical (figure taken from Lange et.al) …………………...19
Figure 13: Layout and Experimental Setup……………………………………………………..23
Figure 14: Mixing Curves comparing different flowrates at 700 rpm agitation…...……………28
Figure 15: Effect of agitation on concentration curves………………………………………….29
vii
Figure 16: Calculation of effective residence time from tracer data……………………………30
Figure 17: Visualization of space time for steady state…………………………………………34
Figure 18: ln Rate vs ln Cg for constant pH…………………………………………………….42
Figure 19: ln Rate vs ln H+ for constant Glucose loading……………………………………...44
Figure 20: Arrhenius plot (Glucose) …………………………………………………………....46
Figure 21: ln Rate vs ln Cf………………………………………………………………………49
Figure 22: ln Rate vs ln H+ (Fructose)………………………………………………………….50
Figure 23: Arrhenius plot (Fructose)……………………………………………………………51
Figure 24: ln Rate vs Ln Cx……………………………………………………………………..52
Figure 25: Arrhenius Plot (Xylose)……………………………………………………………..53
Figure 26: Comparison of sugar dehydration rates based on rate constant……………………..55
viii
Tables
Table 1: Effect of Temperature on Density………………………………………………..……38
Table 2: Maximum difference in rates calculated by accounting for change in effective residence
time………………………………………………………………………………………………38
Table 3: Effect of temperature on proton concentration………………………………………...40
Table 4: HMF formation rates from glucose feed solution……………………………………...41
Table 5: Lumped Rate Constants for glucose experiments (Constant H+ conc.)……………….43
Table 6: Reaction orders for constant pH data (Glucose experiments)…………………………43
Table 7: Lumped Rate Constant k” for constant Glucose loading………………………………45
Table 8: Second order rate constants for glucose calculated from experimental rates…..…...…46
Table 9: HMF formation rates from fructose feed solution (experimental)..…………………...49
Table 10: Second order rate constants for fructose calculated from experimental rates………..50
Table 11: Rates of furfural formation from xylose feed solution………….……………………53
Table 12: Second order rate constants for xylose calculated from experimental rates………….53
Table 13: Rate constants for all sugars at fixed temperatures……………….…………………..54
1
Chapter I
Introduction
Crude oil has been the main feedstock for most of the chemical industry since the early
19th century. “Building block chemicals” in most chemical processing industries can be traced
back to crude oil. An examination of products manufactured by way of petrochemical processing
reveal our dependence on crude oil. Plastics and polymers, paints, fuels, reagents and chemical
solvents are all petrochemical products. Although energy supplies may be supplemented through
the use of solar power, wind turbines and geothermal sources, it is established that our only
renewable source of carbon for chemical manufacture is biomass. Several sources and studies
find that society can depend on the reserves of crude oil for several decades to come, however,
the environmental repercussions of such dependence and the increasing demands of
technological advances and population growth make current consumption patterns unsustainable
[1-5]. New techniques of sourcing crude oil from “fracking” and subsequent processes have taken
their toll on the environment. Several studies indicate that the use of biomass-derived fuels
directly lessen emission impacts. Also, their large scale manufacture has a relatively smaller
impact [6-9]. Although short term fuel price fluctuations may not indicate a long term shortage,
depletion of crude oil reserves is inevitable (projected cost in the year 2025, $54 per barrel)10.
The petrochemical industry was established in the early 20th century and has evolved to
become very efficient and economical with regards to overall yields and conversions. We can
make use of existing chemical engineering technologies (separation, hydrolysis, dehydration and
pyrolysis, among others) to modify the existing refinery setup to operate on a different feedstock
(Biomass Refinery) 10.
Selected biomass feeds may be used in the manufacture of specialty chemicals,
pharmaceuticals, natural polymers and other higher value products. When we consider fuel
2
manufacture from biomass, an important constraint that needs to be managed is the heat content.
Increasing the energy density is the primary focus in fuel technology. This requires the removal
of oxygen as carbon dioxide and water. Some of the oxygen may be left behind in the interest of
better combustion properties. Dehydration and hydrogenolysis are effective strategies used to
reduce oxygen content of biomass12. In general, biomass is found to have approximately 45 wt%
oxygen.
An important factor in changing manufacturing processes is maintaining economic
feasibility of the process. Biomass may be sourced as a virgin biomass or waste biomass. Waste
biomass consists of remnants of harvesting food crops, wastes from paper industry and wastes
from sugar manufacture. Virgin biomass consists of crops or plants grown specifically for use in
biomass processing. Virgin biomass brings into question the food versus fuels argument and so,
efforts have been made to maximize waste biomass use in processing. Bioengineering can help
to manipulate the growth patterns/requirements of these crops. Waste biomass in the form of
landfill/industrial residues, paper pulp black liquor, cornstover may be used as a cheap feedstock,
in some cases negative-cost. Waste energy content in the US amounts to a 15.1 EJ/year which is
significant considering annual energy consumption of 41 EJ/year3.
Experts in the field predict that by the year 2030, 20% of transportation fuels and 25% of
chemicals will come from biomass10.Most of the major petrochemical oil companies have
identified this trend and have begun to develop technologies for biofuel and biochemical
production14-17. So, how does biomass figure into the scheme of processing steps? To understand
this we need to take a closer look first at the specific components of biomass.
3
Chapter II
Literature Review
2.1 Biomass
Figure 1: Composition of Biomass
4
All types of biomass originate from plant sources. Fixed carbon, light and water are
converted to sugar in the photosynthesis reaction. Plant sugars are stored within plant cells in the
form of sugar polymers. The nature and distribution of these polymers within the plant cell is a
function of plant type. Bio engineering has been instrumental in manipulating this distribution to
suit our requirements. Through gene manipulation, scientists have successfully increased
cellulose distribution by reducing lignin in plant species while maintaining plant growth
advantages45.
Figure 1 illustrates the distribution of different chemical species present in biomass.
Biomass is primarily composed of 75-90% sugar polymers, 25-10% non-sugar polymers and
approximately 1% triglycerides, sterols, pigments, alkaloids and other compounds. Biomass
processing begins at the sugar polymers. Broadly speaking, sugar polymers may be present in the
form of starch, cellulose or as hemicellulose. Cellulose is the most abundant biopolymer. It is a
polymer of glucose with β- 1, 4 glycoside linkages5. The molecular weight of cellulose can range
from 300,000 to 500,000 g/mol. This type of linkage renders cellulose crystalline. Glucose
oligomers (dimers, trimers, tetramers etc.) are formed during partial hydrolysis of cellulose46.
Glucose monomers are produced through complete hydrolysis of cellulose.47 Hydrolysis of
cellulose is relatively difficult owing to its crystallinity.
5
Cellulose
Figure 2: Structure of Cellulose
Hemicellulose is another branched sugar polymer. Depending on species, it consists of a
branched glucose or xylose substituted with arabinose, xylose, galactose, fucose, mannose,
glucose or gluconic acid. Side chains may also contain acetyl groups or ferulate61. Sugars in
hemicellulose may contain either 5 carbon atoms or 6 carbon atoms. Xylose and arabinose are
examples of the 5 carbon sugars while glucose, galactose and maltose are some 6-carbon sugars.
In general, hemicelluloses, are easier to hydrolyze owing to amorphous nature.
Hemicellulose
Figure 3: Components of Hemicellulose
Starch is made up of Amylose (10-20%) and Amylopectin (80-90%). Amylose and
amylopectin are polymers of glucose. Amylose is water soluble due to its α-1, 4 glycoside
linkages while Amylopectin is insoluble due to its combination of α-1, 4 glycoside linkage and
α-1, 6 glycoside linkage18.
6
1
4
Amylose
1
4 1
6
Amylopectin
4
6
6
1
1
4
4 1
4 1
Figure 4: Components of Starch (Amylose and Amylopectin)
The monomers that form the fundamental building blocks of the sugar polymers may be
classified into pentoses (5 carbon sugar monomer) or hexoses (6 carbon sugar monomers). Also,
they may be functionally different (keto or aldehyde groups). Dehydration of these monomers
produce the chemical molecules of interest to us. The reactions and products that these molecules
generate are different. Pentose sugars produce furfural upon dehydration, while hexose sugars
produce 5-hydroxymethyl furfural.
7
HMF
Figure 5: Structure of Furanic Aldehydes (Furfural and HMF)
The cellulose and hemicellulose found in plant cells are bound by lignin, they are
collectively referred to as lignocellulose. ∼ 2 × 1011 Metric tons of lignocellulose are available
for processing19. Although lignocellulose does contain sugar polymers such as cellulose and
hemicellulose, separating these polymers from the lignin is an energy intensive process and
requires preprocessing steps20. Difficulty in processing lignocellulose arises on account of the
crystalline nature of cellulose and protection by lignin. However, preprocessing of lignocellulose
to cellulose and hemicellulose solves the age old “food vs fuel” argument51. Most remnants of
food processing, such as cornstover may pass through preprocessing steps to yield useable
products. Cornstover has a distribution of 27-48% cellulose, 13-17% hemicellulose 18-25%
lignin20. In this case, the question of cultivating crops solely for fuel doesn’t arise. At the same
time, the uniform quality and desirable properties of feedstocks are maintained.
The non-sugar polymers are comprised of lignin. Lignin is a cross-linked phenolic
polymer, which on de-polymerization yields aromatics. These compounds have a high heating
value and are of a great economic significance. Research into effective lignin processing steps is
underway [22-25].
8
Figure 6: Structure of monolignol units
Dumesic found that a solvent derived from lignin could replace the organic phase
solvents used in biphasic biomass reaction systems. Furanics (5-HMF, furfural) and lactones
(GVL) could be extracted into the lignin derived phenolic solvent to make the process more
sustainable.26
Three main routes may be taken to manufacture fuel compounds from biomass. This can
be illustrated through the use of the following diagram.
9
Figure 7: Biomass Processing/Upgrading routes
Gasification of biomass yields light fractions mostly with larger vapor pressures such as
carbon monoxide, carbon dioxide, hydrogen, methane, trace amounts of higher hydrocarbons
Upgrading to platform
chemicals
10
such as ethane and ethene, water, nitrogen along with a few impurities depending upon the
gasification methods. This stream may be used to produce methanol or in Fischer-Tropsch
reactions to generate longer chain compounds that are of greater economic value. Pyrolysis and
hydrolysis of biomass yield liquid fractions. Of these, pyrolysis yields heavier oils (bio-oil,
pyrolysis oil), tars, acids and aromatics. These are then upgraded or processed to fuels.
Hydrolysis is used to convert the sugar polymers into monomers of sugar. Hydrolysis may be
carried out through the use of either microbial fermentation or acid reaction steps. After
complete hydrolysis, we are left with monomers such as glucose, among the 6 carbon molecules
and xylose, among the 5 carbon molecules.
Figure 8: Sugar monomers Glucose, Fructose and Xylose
11
Figure 9: Fischer projections of open chain and closed chain forms of glucose.
Polyols exist in two forms, open chain form as shown on the left of Figure 9 and closed
chain forms as illustrated on the right. When dissolved in water, there exists a temperature
equilibrium dependent ratio of these two forms of the compound. Glucose, fructose and xylose
exist in similar anomeric forms. Glucose and fructose solutions retain < 1% open chain or acyclic
12
form. Glucose solutions are found to contain a smaller ratio of acyclic to cyclic chain ratios than
fructose 67.
2.2 Upgrading Technologies
Once the simple sugars are produced, several conversion technologies may be used for
further processing (upgrading). Some of these conversion technologies used are biological,
thermal, chemical/catalytic upgrading or a combination of these technologies. Fermentation has
been employed to convert coal and petroleum gas streams into ethanol13. Also, use of
fermentation in processing biomass is an important application of microbial enzymatic
technology. Sugarcane and other virgin biomass species are being used widely to produce
ethanol from starches. Studies are being conducted on the use of lignocellulosic agricultural
wastes as a feed in the fermentative process. For this reason, strains of bacteria that are adept at
fermenting both glucose and xylose need to be studied [55, 56]. Also, since fermentation steps
generally begin at simple sugar monomers, it is imperative to research combining enzymatic
techniques of hydrolysis of starches (cellulose) and enzymatic fermentation. Fermentation quite
often requires preprocessing steps to free the fermentable sugars. These preprocessing steps
include ammonia explosion, aqueous ammonia recycle, controlled pH, dilute acid, flow through,
and lime58. Glucose fermentation can be used to produce a range of products through
recombinant or non-recombinant DNA technologies and through different pathways. Some of
these products are mixed acids (acetic acid, succinic acid, propionic acid, butyric acid, lactic
acid), alcohols (ethanol, butanol, isopropanol) and glycols with carbon dioxide. Each of these
products can be used to manufacture useful products. Succinic acid for example can be used to
13
manufacture pharmaceuticals and biodegradable polymers. Succinic acid is a precursor to many
industrially important chemicals such as adipic acid, 1, 4-butanediol, tetrahydrofuran (THF), N-
methyl pyrrolidinone, 2-pyrrolidinone, gamma butyrolactone and succinate salts. Fermentation
of waste biomass can be used to produce other products such as methane gas (biogas
plants/waste sludge treatment) and hydrogen gas. Hydrogen gas can be used as a prospective
green fuel through the use of fuel cells thereby producing water as a byproduct. Although these
technologies of fermentation to produce hydrogen are only in the fledgling stages of
development, certain anaerobic photosynthetic fermentative strains of bacteria have shown
promise for their H2 producing capability from glucose57. Fermentation does come with its share
of limitations. In most cases, the fermentation microbes are mixed into the feed. Separation of
the suspended microbes and the products of fermentation can be energy intensive. Use of
immobilized growing species on substrates becomes extremely challenging. Dilute feed streams
are required in most cases. Auxiliary batch operations are needed in some cases for
yeast/bacteria cultivation. Once the hydrolysis step is achieved either enzymatically or through
the use of acids, simple sugars are produced and made ready for the next step of the process.
Assuming complete hydrolysis, simple sugars produced may be upgraded to platform chemicals.
Renewable energy research has been focused on ways of breaking down organic wastes
into fundamental molecules and taking advantage of the functional properties to create raw
material chemicals. This is the concept of platform chemicals. Several platform chemicals have
been identified as useful building blocks for chemicals. These resulting compounds may be used
as is, or in subsequent platform chemical manufacture. 5-HMF and furfural have been identified
as platform chemicals. Dehydration of glucose, fructose and xylose to produce platform
chemicals will be studied in the proceeding part of the study.
14
2.3 Acid Catalyzed Dehydration
Hexose sugars on triple dehydration yield HMF. The exact mechanism varies with
regard to type of solvent used [32-35]. No definitive conclusion has been made on mechanism of
hexose dehydration. Isomerization of glucose to fructose and vice-versa is a common factor
between most studies. Side reactions form humins and organic acids like levulinic acid [36, 37].
Acids formed as a result of side reactions have been known to self-catalyze the reaction. Side
reactions are affected by the solvent type, temperature and catalyst. Dehydration products are
one of the four reaction products formed during decomposition of polyols. Isomerization,
fragmentation and condensation also take place at elevated temperatures. Two mechanism
schemes were proposed to explain dehydration of fructose and glucose. The first scheme
suggests a series of reactions commencing with and retaining the fructofuranose ring or cyclic
form of fructose. The second scheme postulates a succession of reactions proceeding mainly via
open-chain or linear intermediates. Literature is in favor of the first scheme. Higher yields of
HMF have been reported in dehydration studies of fructose [62-64]. Rates of HMF production and
also selectivity was higher for fructose dehydration when compared to dehydration of glucose.
Glucose is much cheaper and more widely available when compared to fructose. The catalytic
pathway of glucose and fructose dehydration is shown in Figure 10. Glucose isomerizes to
fructose or to form HMF.
15
Figure 10: Postulated mechanism for the production of hydroxymethylfurfural (HMF) 68.
Biomass derived feed-stocks have low thermal stabilities and their functional groups are
hydrophilic compared to their petrochemical counterparts. This makes aqueous phase processing
a requirement in most of the reactions involving biomass. Early research into sugar chemistry
16
involved detailed models for kinetic studies of hydrolysis of starch29. These studies revealed that
hydrolysis was accompanied by an undesired decomposition reaction. The product was HMF.
Levulinic acid was formed as a result of acid catalyzed degradation of starch. It was only later
realized that these products could be used in processing steps further to yield useful products.
These reactions form an essential step in biomass processing. Dehydration may also be coupled
with hydrogenation to carry out a set of reactions. These reactions form the basis of the biomass
refinery. Aqueous phase dehydrations using mineral acids have been well documented and
studied [3, 27, 28, 30]. Furanic aldehydes (HMF and furfural) may be upgraded to higher alkanes
through a series of condensation reactions and hydrogenation/dehydration reactions31.
17
2.4 Platform Chemicals
2.4.1 HMF
Figure 11: HMF as a platform chemical
HMF can be processed through different chemical processing steps to manufacture
useful chemicals. The oxidation route produces dicarboxylic acid groups such as furan
dicarboxylic acid FDCA which can be used to make products similar to the petrochemical
product PET (poly ethylene terephthalate). Reduction of HMF produces dihydroxymethyl
tetrahydrofuran DHMTHF which can be used to produce polymers as well as in the fuel industry
as already explained in biomass refinery setup 38. Hydrogenolysis of HMF also helps in fuel
manufacture. Molecules such as dimethylfuran (DMF) are useful as fuel additives [39, 40].
Etherification of HMF yields ethers such as 5–alkoxymethylfurfuralether a useful intermediate in
fuel industries [41, 42, 43]. The heterocyclic structure of furans is useful in manufacture of
POLYMERS
Diols/DHMTH
F
FUELS
18
biologically active compounds is an important intermediate for pharmaceutical industry. A series
of hydrogenation, condensation and dehydration reactions may be employed to convert HMF
into linear chain alkanes. Hydrogenation reactions help to saturate all the bonds in the
compound. Aldol condensations increase the overall chain length and the final step in alkane
production is a hydrogenation accompanied by the removal of water molecules to finally produce
the required linear alkane. These series or reactions are manipulated depending upon the desired
chain length of final alkane. It is possible to produce 7-15 carbon alkanes through combined use
of HMF and furfural31.
19
2.4.2 Furfural
Figure 12: Furfural as a platform chemical 11.
Furfural is being researched for a range of specialized chemical applications. Bisphenol-A-
furfural a furfural based polymer resin has shown advantages over phenol-formaldehyde resins
as regards thermal, chemical, physio-mechanical, and electrical properties59. Furfural may be
used to manufacture solvents (methylfuran), furfuryl alcohol and furoic acid. Furfural and
20
hydrogen when fed through Ni or Ni-Fe SiO2 bed, form furfuryl alcohol and furans.
Hydrogenation reactions result in the formation of furfuryl alcohol which then may be
hydrogenolyzed to yield 2-methyl furan. The furans produced in the Ni-Fe catalyzed bed may be
used to produce 4 carbon compounds such as butanal, butanol and butane60. Hydrogenation steps
to upgrade furfural into fuel compounds is still considered to be the more important conversions.
Similar sequence steps used to convert HMF to linear alkanes may be used to convert furfural
into desired alkanes. Special types of reactors are used for these reactions. These reactors consist
of four phases. The aqueous inlet stream with water soluble organic content, the hexadecane
solvent stream, a gas inlet H2 and the solid catalyst (Pt/SiO2-Al2O3). Since dehydration
hydrogenation reactions take place in this four phase setup. This setup is referred to as a 4-PD/H
reactor system31. Aldol coupling of furfural with levulinic acid can be seen in the top right of
Figure 12.
21
2.5 Reaction Orders and Rate Law
In the experiments carried out, sugars were consumed (disappearing reactant). Rate
depends on temperature and composition. “Rate r, is expressed as the product of rate constant, k,
and concentrations of the reactant,
The rate constant, k, is a function of temperature, catalyst, ionic strength and solvent. For
reacting species A and B
−𝑟 = 𝑘𝐶𝐴𝛼𝐶𝐵
𝛽 Equation 1
The rate law involves the product of rate constant and reacting species concentrations with their
respective exponents. In the case of our reactions of sugar and protons, we have,
−𝑟 = 𝑘𝐶𝑠𝑢𝑔𝑎𝑟𝛼 𝐶
[𝐻+]
𝛽 Equation 2
Rate data collected can be used to determine the activation energy and pre-exponential factors
for the reaction. To be able to do so, we use the Arrhenius Equation.
𝑘(𝑇) = 𝐴𝑒−𝐸𝑎/𝑅𝑇 Equation 3
The Arrhenius plot is a great way to extrapolate rate coefficient data for temperatures that have
not been investigated. 𝐴 in the above equation represents the pre-exponential factor, 𝐸𝐴 is the
activation energy, R is the gas constant (8.314 J/mol) and Temperature “T” is measured in
degree Kelvin.
This study aims to calculate the rates, activation energy and pre-exponential factors for
reactions with different sugars. By doing this, it is also possible to compare different catalysts for
similar dehydration reactions. Just as increasing temperatures increase rate constants, changing
catalysts may alter the nature and number of molecular collisions and increase rates. Reactions
carried out at different temperatures yield different values of rate constant k. Plotting ln k as a
22
function of (1/T) yields a slope of –Ea/R. The Arrhenius equation may also be written in semi-
log form as,
ln 𝑘 = 𝑙𝑛𝐴 −𝐸𝑎
𝑅𝑇 {i.e. y= b + mx } Equation 4
23
Chapter III
Layout of the CSTR and Design
Figure 13: Layout and Experimental Setup
The setup comprised of (upstream to downstream) the feed reservoir, an HPLC pump, a
relief valve rated at 1000 psi (well above operating values and below operating limits of
equipment), a back pressure regulator (BPR) rated at 500 psi, the insulated CSTR, a condenser,
and another 500 psi rated back pressure regulator. The pressure gauge positioned above the
CSTR was used to monitor pressure during the experimental runs. The CSTR was placed
between the two BPR’s in the interest of maintaining reaction pressures. Two thermocouples
were used to monitor temperatures, one thermocouple was located in the stainless steel shell
closest to the rod heating elements of the CSTR to measure highest attainable temperatures of the
system while the other thermocouple was positioned at the top of the reactor and used to measure
actual reacting fluid temperatures. Two thermocouples were used in the interest of maintaining
an acceptable temperature gradient without exceeding the temperature tolerance of the liner
material. The temperature set point for the controller was based on the temperature readings from
24
the reactor shell thermocouple. A BPR malfunction could cause a spiraling effect and disrupt all
readings. A blocked BPR resulting from polymerized humin formation could cause pressure
buildup in the CSTR. The reduction in flow and consequent reduction of the heat sink would
result in a temperature increase. This situation would imply waiting through another temperature
transient. An acetone wash, sonication and pumping acetone through the BPR for 10 minutes
solved flow issues by dissolving the humins blocking flow. Temperature was also found to alter
the mechanical properties of the BPR. Operating at higher temperatures resulted in a pressure
spike. The ice-bath condenser served to solve this problem and allow for uniformity in readings.
The condenser served to stop the reaction and volatilization of reacting solution in the tubing
downstream to the CSTR. The CSTR used was constructed out of stainless steel 316 owing to its
mechanical properties and ease of fabrication. Despite its resistance to corrosion due to the
nature of experiments to be run (low pH and high temperature) it was decided to use a lining
material Polyether ether ketone (PEEK) on the inside of the working reactor so as to eliminate all
contact of the reacting fluid with any metallic surface. Leaching of metal shell contaminants
could be minimized by eliminating all metal contact with the fluid stream. The choice of solvent
was water (discussed in previous sections). The inside pressure was to be regulated above the
vapor pressure of water. All these efforts were taken to ensure homogeneous operation in the
liquid phase. The insulated reactor was positioned on the magnetic pad such that the rod stirrer
could attain desired mixing. The newly fabricated CSTR was seal tested using nitrogen gas at
pressures above rated operating values. A feed inlet dip-tube was used to prevent back-mixing.
Before beginning dehydration experiments it was necessary to make sure the mixing in the
CSTR was near ideal. Several methods can be used to study and understand mixing
characteristics, these will be highlighted further.
25
Chapter IV
Mixing
For the purpose of good mixing, several agitators were tested. The mixing characteristics
were studied from tracer experiments to understand the limitations of mixing and corrective
measures were employed by trying out different agitation speeds ranging from 700 rpm to 1200
rpm. When mixing tends to diverge from ideal conditions, well mixed models are inadequate.
During the development of the reactor, mixing conditions were unknown. We make use of
changing concentrations to understand the nature of mixing in the reactor. By using tracer
experiments, either pulse inputs or step inputs, it is possible to compare data to ideal tracer
curves and diagnose mixing problems. Specific tracer tests are needed to run mixing diagnostics.
In all cases, the output of tracers are measured as a function of time and input concentrations.
4.1 Experimental Methods (Mixing)
For the purposes of this experiment we choose a step input for convenience. For the
tracer test, we require an inert molecule which is easily detectable and inexpensive. The tracer
must also possess similar properties as the reacting fluid to be tested. Inert tracers such as salt
solutions are easy to work with. Also, it is inexpensive and easily detected by an in line total
dissolved solids (TDS) meter. This allowed for an immediate output response to flowing feed
stream.
The inline TDS meter was operated using two probes, one for measuring feed
concentrations and the other for the CSTR output. Several tracer tests were performed, many of
the charts comprise data that are of no consequence to the final reactor since several
26
modifications were applied. However we can get a good idea of what prompted some of these
changes, and the identification of mixing problems.
First, different flow rates were tried out to visualize what information could be extracted
from the mixing curves. The plots are concentration versus time and the curves are called “C”
curves. Concentrations are normalized against the feed concentration since it is the maximum
concentration attainable. We intend to see how much time it takes for the concentration of the
exit stream to reach the desired “1” on the y-axis. Also, the extent of deviation between ideal and
theoretical curve may be analyzed. From the data collected, it was realized that a few minutes
elapsed before the output stream began to show any signs of the input tracer fluid. All readings
were normalized with respect to the input concentration. For all analyses, output response was fit
to the response to step input equation for CSTR
𝐶𝑡 = 𝐶𝑡0(1 − 𝑒−
𝑡
𝜏) Equation 5
For the condition of 1 ml/min in Figure 14, first, the reactor was readied by setting an agitator
speed of 700 rpm. The tracer input was then started and the output readings were noted in minute
increments. Since the concentrations were normalized, i.e. C/C0. The ideal curve based on
Equation 6 was compared to experimental data:
𝐶𝑡
𝐶𝑡0
= 1 − 𝑒−𝑡
𝜏 Where t is in minutes. Equation 6
Nominal residence time τ was calculated for a measured volume of 13.66ml. Nominal residence
time τ = 13.66 min for the first condition. (Since τ =𝑉𝑜𝑙𝑢𝑚𝑒 𝑜𝑓 𝑟𝑒𝑎𝑐𝑡𝑜𝑟
𝐹𝑙𝑜𝑤 𝑟𝑎𝑡𝑒)
For the second condition, the only change made was the flow rate. Flow was increased to 2
ml/min and the residence times changed to 6.5min. For the third condition, 3ml/min, τ=4.5 min.
27
For the HPLC analysis, a solution was prepared with a known amount of glucose, levulinic acid
and HMF. The stock solution response area obtained from the HPLC was noted as C0. Samples
were collected after 5 minute intervals and analyzed in the HPLC. The response area was taken
as C. The HPLC confirmed that the data collected using the inline TDS meter was accurate and
sufficient for the mixing experiments.
4.2 Mixing Results and Discussion:
In the first mixing analysis, ideal and experimental mixing curves were plotted side by side
with the intention of comparing flowrate effects. Through Figure 14 it was made clear that for a
given agitation speed, flowrates impact time taken to reach maximum output concentrations.
Since residence times are lowered with increased flowrates, maximum concentrations were
reached sooner. Curves representing 3 ml/min flowrates attained maximum concentration around
the 20 minute mark followed by the 2 ml/min which attained maximum concentrations around
the 35 minute mark. Curves representing 1 ml/min could not reach maximum concentrations
within 45 minutes of operation.
28
Figure 14: Mixing Curves comparing different flowrates at 700 rpm agitation.
Since preliminary analysis showed that maximum exit concentrations could be attained soonest
for higher flow rates. It was decided to run the next test of a 3 ml/min flow rate keeping all other
factors constant.
The bar type agitator was used in all the experiments and was observed to be unstable at higher
agitation speeds. A bar type stirrer was tested between 700 and 1200 rpm. The occurrences of
instability and vibration of the magnetic stirrer increased above 1000 rpm.
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 5 10 15 20 25 30 35 40 45
No
rmal
ized
Co
nce
ntr
atio
n
Time (min)
Experimental 3mlpm Ideal 1mlpmExperimental 2mlpm Ideal 2mlpmIdeal 3mlpm Experimental 1mlpm
29
Figure 15: Effect of agitation on concentration curves. All curves above are for 3 ml/min flow rates.
In the second mixing analysis shown in Figure 15, effect of agitation speeds was studied. The
flow rate was kept constant at 3 ml/min and two tracer tests were performed, one at 700 rpm and
the other at 900 rpm using the same feed solution. Ideal curves were fit to the experimental data.
The influence of simply changing agitation is illustrated in Figure 15. There was a qualitative
increase in mixing without any instability at 900 rpm agitation speeds. The final output
concentrations obtained for the 700 rpm experiment were lower than final output concentrations
seen in the 900 rpm agitation experiments within the 40 minute mark. This comparison
confirmed increased mixing efficiency for higher agitation speeds.
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 5 10 15 20 25 30 35 40 45
No
rmal
ized
Co
nce
ntr
atio
n
Time (min)
Curve fit for 700 rpm data
Curve fit for 900 rpm data
ideal
30
Figure 16: Calculation of effective residence time from tracer data. Experimental data collected at 900
rpm agitation and 1ml/min flow rate.
Nominal residence times would be insufficient to accurately calculate rates owing to possible
dead volume. The measured volume of the CSTR was 13.66 ml. Experimental data collected for
the final curve are shown in Figure 16. The curve used to fit the experimental data represents a
residence time of 14.60 min. This is our effective residence time for a flow rate of 1 ml/min and
agitation of 900 rpm. A larger effective residence time calculated would normally indicate that
the measured volume of the reactor failed to account for the dead volume relative to flow rate. In
the case of our experimental setup, pump flowrates were found to deviate from 1 ml/min. This
could explain higher effective residence times as confirmed in Figure 16.
Although higher flow-rates were desirable for reaching maximum concentrations sooner,
1ml/min was chosen as the set operating volumetric flow-rate. Larger flow rates would translate
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 10 20 30 40 50 60 70 80
No
rmal
ized
Ou
tpu
t C
on
cen
trat
ion
Time (min)
Experimental Output
Curve fit to Effective Residence Time
31
to a larger heat sink. Maintaining high temperatures with higher flow rates could result in large
thermal gradients sometimes exceeding the operating temperature specifications of the liner
material.
32
Chapter V
Dehydration Experiments
5.1 Analytical Equipment and Chemicals
All dehydrations were carried out in the CSTR setup described above using sulfuric acid
as the catalyst. The chemicals obtained from all sources were used as is without any additional
purification. Alpha-D-(+)-glucose, 99+% and sulfuric acid 95% were purchased from ACROS
Organics, D-Fructose lab grade from Fischer Scientific, D-(+)-xylose used in experiments was
sourced from Sigma Aldrich. K type thermocouples (OMEGA) were used for the temperature
measurements. Cartridge heaters were used to heat the reactor system. Analysis was carried out
in an HPLC (Hi-Plex H+ column Agilent Technologies) using the sugars PLEX column. Sugars
were analyzed using the refractive index peaks and dehydration products were measured using
ultra violet (210 nm) peaks. A sulfuric acid solution with a pH of 2 was used as the mobile
phase.
5.2 Selection of Reactor
A preliminary study was conducted to understand the nature of reaction and chemistry,
industrial applications, text book methods and formulae. Several types of reactors may be used
for a dehydration reaction. Some research is focused on batch reactors for dehydrations in a
biphasic setting. The Biofine process (an industrial scale setup for production of HMF, levulinic
Acid and furfural) combines hydrolysis and dehydration in a plug reactor and a CSTR. The
choice of reactor type is based on specific goals of the study, economy, practicality and basic
functioning. Our need was based on running liquid phase reactions. The decision of using a
33
CSTR over other reactor types was based on several factors. A CSTR can be used for uniform
products and continuous production. The temperature could be controlled and distributed
uniformly due to the well mixed nature of a CSTR. Near isothermal conditions were needed in
the CSTR to be able to precisely maintain and report reaction temperatures. Many aspects of the
CSTR were planned, designed and controlled to make this possible. Calculations used to
determine rate constants and other parameters in the proceeding part of the text rely on accurate
recorded measurements of temperature. Specific rate data could be collected at several points for
a given set of conditions, thereby increasing the accuracy of data. Mixing transients could be
well documented through tracer tests and accurate steady state data could be collected for a given
reaction. Data acquired at steady state was a crucial factor in the kinetic study. Steady states were
important because data for rate constants needed to be calculated for a fixed temperature
(temperature steady state). Steady state in concentrations also needed to be maintained to
understand the influence of specific concentrations of protons and sugars in feed.
We choose homogeneous liquid phase reactions due to the added advantages. There
are no transport limitations experienced as are with heterogeneous catalysts. Surface anomalies
do not need to be considered. Other concerns that need to be addressed when using solid
catalysts are catalyst site characterization, poisoning and deactivation. The choice of acid was
based on ease of measurement of acid strength. Sulfuric acid has been studied extensively for use
in dehydration experiments (HMF and furfural production) and its selection as a catalyst would
simplify most experimental techniques [37, 64].
34
5.3 Experimental Procedure
The decision of running the reactor over a range of reactor temperatures (130-150℃)
was based on the literature study and temperature limitations of the CSTR. Carefully prepared
feed solutions at ambient temperatures were checked using a calibrated pH probe (buffers pH
values were 4.01 and 7.01) to ensure that proton concentrations matched calculations based on
measured acid loading. Although acid loading was measured in molar units, an equivalent proton
concentration based on dissociation was developed and used. A feed solution sample was
analyzed against the calibrated HPLC chromatogram data to confirm that the sugar loadings
were within boundaries. The pump was set to 1ml/min. De-ionized water was pumped through
the reactor while the temperature reached the set point. Feed solution was pumped through the
CSTR only after all temperature transients were overcome. End of loading feed solution was
taken as t = 0. The mixing-flow transient for this residence time amounts to approximately 68
minutes Figure 17.
Figure 17: Visualization of space time for steady state. Approximately 5 space times needed for steady
state.
0
0.2
0.4
0.6
0.8
1
1.2
0 1 2 3 4 5 6
No
rmal
ized
Co
nce
ntr
atio
n
Space Time (t/τ)
35
A total of 5 residence times are needed to ensure steady state with regard to flow. Once steady
state was achieved, 4 samples were collected for every experiment to ensure uniformity. The first
sample was collected at t = 80 min. A single sample was collected every 3 minutes to be able to
ensure uniformity and reproducibility in data.
5.4 Analytical Methods
Densities of output solutions after experiment were compared to feed solutions before
experiment. The change in density after completion of experiment were negligible. However,
due to temperature dependence of densities during reaction, feed densities needed to be measured
and recorded. Feed densities were measured manually by weighing out 100 ml of the unused
feed solution. Concentrations of the feed input of an experiment and also output concentrations
were measured from calibrated HPLC chromatogram areas. Concentration data was recorded in
moles per sample volume. Carbon moles were calculated for both input and output. One mole of
a hexose sugar corresponds to 6 moles of carbon. Similarly one mole of a pentose corresponds to
5 moles of carbon. Carbon moles were calculated for all inputs and corresponding outputs. The
difference in measured inputs and outputs helped maintain a carbon balance. The carbon balance
was maintained within a range of 5% error. Large deviations in the carbon balance would be
indicative of a system leak or unaccounted reactant or product. Conversions were calculated
based on the amount of product produced per unit reactant (sugar) fed to the reactor i.e. moles
product per mole reactant fed. Rates of reaction were calculated based on the product.
r =moles produced per volume
residence time Equation 7
36
Rate estimations are affected by changing densities. As temperature increases, fluid in the
reactor expands and its volume increases. This causes a perturbation in residence time relative to
that measured at room temperature. Since residence times are used to estimate rates of reaction
(Equation 7), it is important to correct for temperature-dependent changes in density under
reaction conditions. The volume increase is a function of initial density and depends on
Volumetric Expansion Coefficient (β) with units /℃ and Bulk Modulus (E) with units N/m2. The
combined effect of heat and pressure can be calculated as:
ϱ =ϱ0
(1 + β (𝑡1 – 𝑡0))×(1 –(𝑝1 – 𝑝0)/E) Equation 8
ϱ: Density at operating temperature
ϱ0: Density of feed at room temperature
β: Volumetric expansion coefficient taken as 0.0002 (m3/m3 ℃) (water)
𝑡1 – 𝑡0: Difference in final temperature and room temperature (℃)
𝑝1 – 𝑝0: Difference between ambient pressure and reactor pressure at operating
temperature (Pa) fixed for all reactions at 33.5 × 105 Pa
E: Bulk Modulus taken as 2.2 × 109 N/m2
Proton concentrations of reacting solutions were calculated by solving equilibrium equations.
𝐻2𝑆𝑂4 ↔ 𝐻+ + 𝐻𝑆𝑂4− Equation 9
𝐻𝑆𝑂4− ↔ 𝐻+ + 𝑆𝑂4
− Equation 10
𝐻2𝑂 ↔ 𝐻+ + 𝑂𝐻− Equation11
37
𝑘𝑎1 =[𝐻+][𝐻𝑆𝑂4
−]
[𝐻2𝑆𝑂4] Equation 12
𝑘𝑎2 =[𝐻+][𝑆𝑂4
−]
[𝐻𝑆𝑂4−]
Equation 13
𝑘𝑤 =[𝐻+][𝑂𝐻−]
[𝐻2𝑂] Equation 14
Calculation of dissociation constant of water kw is as under
ln(𝑘𝑤) = 2.70291 − 2.6105 × 103𝑇 (Temperature T in degree Celsius) Equation 15
Sulfuric acid is a diprotic acid, i.e. It contains two protons. Sulfuric acid may dissociate to donate
one proton or it may donate the second proton as well. The first dissociation is a strong acid
dissociation. The dissociation constant ka1 for the first dissociation is large of the order of
2.4×106. We may assume complete dissociation of the first proton due to the high value of ka1.
Dissociation of sulfuric acid in water is exothermic. We know that the proton concentration of
the solution will decrease with an increase in temperature. We need to quantify this decreased
proton concentration by solving equilibrium reactions and considering all of the temperature
dependent terms. The second dissociation constant ka2 (temperature dependent) is a weak acid
dissociation and is known for the range of operating conditions. It may be calculated using the
Debye-Huckle equations for ionic strength.
log 𝐾2 = 56.889 − 19.8858 𝑙𝑜𝑔 𝑇 −2307.9
𝑇− 0.006473𝑇 (𝑇𝑒𝑚𝑝𝑒𝑟𝑎𝑡𝑢𝑟𝑒𝑠 𝑇 𝑖𝑛 𝐾𝑒𝑙𝑣𝑖𝑛)
Equation 15
38
5.5 Kinetic Results and Discussion
5.5.1 Glucose
Before rate data were collected, corrections were applied to account for the change in
density and the change in proton concentration of reacting solution. The effect of temperature on
density was studied and illustrated in Table 1 by evaluating the percentage change in density for
the highest and lowest temperatures within operating temperature range. Residence times were
affected as a result of this change in effective volumetric flow-rate. Rates were found to vary by
up to 4.27% for higher temperatures as can be seen in Table 2. Change in densities of reacting
solutions after exiting the CSTR were negligible and not considered owing to small conversions.
TABLE 1: EFFECT OF TEMPERATURE ON DENSITY.
Operating Temperature ℃
Initial density at 25℃ (kg/l)
Final density at operating temp (kg/l)
% Change
150
1040 1016 2.3
1014 990 2.3
1004 981 2.3
1001 978 2.3
120
1040 1022 1.7
1014 996 1.7
1004 986 1.7
1001 983 1.7
TABLE 2: MAXIMUM DIFFERENCE IN RATES CALCULATED BY ACCOUNTING FOR CHANGE IN EFFECTIVE RESIDENCE TIME.
Temp ℃
Glucose wt %
pH measured
with probe
Rate HMF (mol/l.min)
Volumetric flow rate (ml/min)
Residence time at
25℃
Rate not accounting for temp.
Density at
141℃
Effective flow-rate
Residence time at 141℃
Rate at 141℃
Difference in calculated
rates
141 1 2.01 1.678E-07 0.959 14.23 1.609E-07 979.79 0.979 13.64 1.7E-07 4.27%
39
Most mechanistic studies neglect the effect of proton concentrations. They instead rely on
rate measurements for specific pH values 52. The activation energies and pre-exponential factors
calculated in these studies lump the effects of proton concentrations. This is possible through the
use of the modified Saeman equation which is similar to the Arrhenius equation. This equation
can be expressed as;
𝑘 = 𝐴0. [𝐻]𝑚𝑒−𝐸𝑎𝑅𝑇 Equation 16
Where, 𝐴0 is the Saeman pre-exponential factor, H is the proton concentration and m indicates
temperature dependence of H.
In this project, we focus on calculating proton concentrations at reaction temperatures
through the use of measured proton concentrations at room temperature. Dissociation is
temperature dependent and changes for every temperature. Molar concentrations of protons
could not be measured during the reaction and therefore called for calculated values. The
inclusion of proton concentrations at reaction temperature are needed for calculation of rate
constants. Our plot of the Arrhenius equation in Figure 20 includes the effect of proton
concentrations separately and does not require normalizing the pre-exponential factor with
proton concentrations. We avoid the use of lumped pre-exponential factors and hence are able to
express partial orders in both sugars and protons.
Table 3 illustrates recorded values of pH and H+ at room temperature with
corresponding pH and H+ values at reaction temperatures. The temperature column represents
experimental reaction temperatures. The difference in the proton concentrations at room
temperature and at reaction temperatures may be represented as a percentage as shown in the last
column of Table 3.
40
Largest changes in proton concentration could be observed for higher temperatures and higher
pH values. Proton concentrations were found to vary with temperature by up to 45.9% for higher
temperatures of 141℃ and pH 2.
TABLE 3: EFFECT OF TEMPERATURE ON PROTON CONCENTRATION.
Temperature ℃
pH @25
℃
H+ @ 25℃
(mol/l)
pH @ Final Temperature
H+ @ Final Temperature
(mol/l)
% difference in H+ due to heating
123.00 1.00 0.099 1.037 0.091 8.57
131.80 1.50 0.031 1.592 0.025 22.36 132.00 2.00 0.008 2.234 0.005 43.45 132.50 1.00 0.099 1.038 0.091 8.77
141.00 2.01 0.008 2.241 0.005 45.90
141.00 1.50 0.031 1.594 0.025 22.95
141.00 1.00 0.099 1.038 0.091 8.92
103.60 2.00 0.008 2.201 0.006 33.05
122.00 2.00 0.008 2.226 0.005 40.59
123.00 2.07 0.008 2.243 0.005 40.93
123.00 1.50 0.031 1.590 0.025 21.65
Rates of dehydration of glucose were calculated based on moles of HMF produced per unit
volume per unit time. The corrected residence times contributed to reliable rate measurements
illustrated in Table.4.
41
TABLE 4: HMF FORMATION RATES FROM GLUCOSE FEED SOLUTION. (HMF YIELDS RANGING FROM 0.26 ΜMOL/L FOR PH
2, 1 WT % GLUCOSE TO 226.02 ΜMOL/L FOR PH 1, 10 WT % GLUCOSE)
Rate data were collected at constant temperature. From the rate law, we have;
𝑟 = 𝑘𝐶𝑔𝑎𝐶𝐻+
𝑏 Equation 17
Rate data were calculated at several known concentrations of sugar and protons. We are
able to extract information from the specific groups of data points. By grouping the rates
measured at fixed pH (proton concentrations) and fixed sugar concentrations it is possible to
determine partial orders using kinetic analysis.
Rate data collected for a constant pH by varying sugar concentrations helps express the
effect of sugar concentrations at that temperature. In the case of this experiment, proton
concentrations are fairly unchanged during reaction since protons play the role of a catalyst. For
constant pH, the rate law may be represented as;
123℃ 132℃ 141℃
Glucose (wt %)
pH Rate (mol/l.min)
Glucose (wt %)
pH Rate (mol/l.min)
Glucose (wt %)
pH Rate (mol/l.min)
1
2 1.87E-08
1
2 5.83E-08
1
2 1.67E-07
1.5 6.29E-08 1.5 1.88E-07 1.5 5.52E-07
1 2.28E-07 1 7.98E-07 1 1.77E-06
2
2 3.76E-08
2
2 1.21E-07
2
2 3.46E-07
1.5 1.25E-07 1.5 3.89E-07 1.5 1.12E-06
1 4.80E-07 1 1.44E-06 1 3.42E-06
5
2 9.42E-08
5
2 2.81E-07
5
2 8.39E-07
1.5 2.95E-07 1.5 9.52E-07 1.5 2.62E-06
1 1.22E-06 1 3.62E-06 1 8.38E-06
10
2 2.05E-07
10
2 5.64E-07 10 `
2 1.61E-06
1.5 5.36E-07 1.5 1.77E-06 1.5 4.82E-06
1 2.04E-06 1 6.69E-06 1 1.53E-05
42
𝑙𝑛𝑟 = 𝑙𝑛𝑘′ + 𝑎𝑙𝑛𝐶𝑔 Equation 18
Lumped rate constant k’ as illustrated in Table 5 may be obtained from the regression plot of
multiple data points.
Figure 18: ln Rate vs ln Cg for constant pH
-18
-17
-16
-15
-14
-13
-12
-11
-3-2.5-2-1.5-1-0.5
Ln R
ate
Ln Cg
pH 2 123degC pH 1 123degC pH 2 132degC
pH 1 132degC pH 2 141degC pH 1 141degC
43
TABLE 5: LUMPED RATE CONSTANTS FOR GLUCOSE EXPERIMENTS (CONSTANT H+ CONC.)
pH Lumped Rate Constant k' (1/min) Temperature ℃
2 3.41E-07
123 1.5 1.08E-06
1 4.55E-06
2 1.02E-06
132 1.5 3.51E-06
1 1.35E-05
2 3.11E-06
141 1.5 9.75E-06
1 3.14E-05
TABLE 6: REACTION ORDERS FOR CONSTANT PH DATA (GLUCOSE EXPERIMENTS)
Temperature ℃ pH Reaction Order
123
1.0 0.945
1.5 1.008
2.0 1.044
132
1.0 0.986
1.5 0.990
2.0 0.997
141
1.0 0.973
1.5 0.957
2.0 0.963
Data in Figure 18 allows us to observe the cumulative set of experimental points for
constant pH. Each line in the graph represents four reactions carried out at a single pH and fixed
temperature by varying glucose concentration alone. It can be determined from Table 6 that
partial order is approximately “1” with respect to glucose.
44
A similar analysis for constant glucose loading may be performed. For constant glucose
concentrations, the rate law may be represented as;
ln 𝑟 = 𝑙𝑛𝑘" + 𝑏𝑙𝑛𝐶𝐻+ Equation 19
Here, k” illustrated in Table 7 are the lumped rate constants.
Figure 19: ln Rate vs ln H+ for constant Glucose loading
-18
-17
-16
-15
-14
-13
-12
-11
-10
-5.5-5-4.5-4-3.5-3-2.5-2
Ln R
ate
Ln H+
1% 123degC 1% 132degC 10% 132degC
1% 141degC 10% 141degC 10% 123degC
45
TABLE 7: LUMPED RATE CONSTANT K” FOR CONSTANT GLUCOSE LOADING
Temperature ℃ Glucose (g/l) Lumped Rate Constant k"
(1/min)
123
10 2.45E-06
20 4.90E-06
50 1.15E-05
100 2.09E-05
132
10 7.39E-06
20 1.53E-05
50 3.73E-05
100 6.95E-05
141
10 2.17E-05
20 4.43E-05
50 0.000103
100 0.00019
Data in Figure 19 allows us to observe the cumulative set of experimental points for
constant glucose concentration. Each trend line represents data collected at a fixed temperature
and glucose concentration for three different pH values. It can be determined that partial order is
approximately one with respect to protons.
First order reactions (partial) in glucose and protons individually are observed in the preceding
sections. Once we have our partial orders, overall order of the reaction may be calculated as
n= a + b Equation 20
From our data it is clear that overall order of reaction is “2”. Since units of k are
𝑚𝑖𝑛−1(𝑚𝑜𝑙
𝑙)[1−(𝑎+𝑏)] Equation 21
We get units of the overall rate constant as 𝑙/(mol.min). We may now use the rate law with
known orders to ascertain the overall rate constant k at each temperature.
46
TABLE 8: SECOND ORDER RATE CONSTANTS FOR GLUCOSE CALCULATED FROM EXPERIMENTAL RATES
Temperature ℃ Rate constant [l/mol.min]
123 4.38E-05
132 0.0001
141 0.0003
Rates measured for all concentrations of glucose and protons are used in all proceeding analyses.
Rate constants were calculated for second order reaction in protons and sugar at every reaction
temperature. The least squares numerical method was used to calculate k.
Calculation of Rate Constant, Activation Energy and Pre-Exponential factor.
The Arrhenius plot or Ln k versus (1/T) graph was used to explore the temperature
dependence of the dehydration reaction.
Figure 20: Arrhenius plot (Glucose)
-12
-11.5
-11
-10.5
-10
-9.5
-9
-8.5
-8
-7.5
0.0024 0.00245 0.0025 0.00255 0.0026
ln K
1/T
47
The slope of this plot corresponds to Ea/R (shown in literature survey). Ea is calculated as
154.5 ± 11.75 kJ/mol. Activation energy value from literature studies found to be 118±37.5
kJ/mol.49 Pre-exponential factor A corresponds to the exponent of the intercept. A is calculated
as 1.115E+16 l/(mol.min).
Similar experiments and analyses were carried out with fructose and xylose.
48
5.5.2 Fructose
Fructose dehydration rates as experimentally determined are illustrated in Table 8.
Fructose dehydration is found to have higher rates than glucose dehydration and are in agreement
with literature. Rate constants were used to compare rates of dehydration of glucose and fructose
and are presented in Table 10. Rates of fructose dehydration were found to be 99% faster than
glucose dehydration. This is in accordance with the cyclic intermediate scheme presented in our
study of mechanism. The alternative theory of reactions proceeding via open chain intermediates
is unable to explain a difference in rates of glucose and fructose dehydrations. The elimination of
the 3-hydroxyl group to form the enol that is thought to take place through the liner chain
mechanism should commence at the same rate for both glucose as well as fructose. However,
this is not the case. Both dehydration reaction schemes lead to the formation of HMF through the
fructofuranosyl intermediate. Cyclic/Ring mechanisms leading to the intermediate formation
through 𝛽-elimination can explain differential rates between glucose and fructose. Anomeric
equilibria and the effects of temperature on tautomerism were not explored in this study.
49
TABLE 9: HMF FORMATION RATES FROM FRUCTOSE FEED SOLUTION (EXPERIMENTAL).
Temp ℃ pH Fructose (mol/l) Rate (mol/l.min)
103.6 2
0.055 1.91E-07
0.110 4.52E-07
0.274 1.13E-06
112.5 1
0.053 1.19E-05
0.257 5.78E-05
122.4 1
0.049 3.01E-05
0.241 0.0001
104.1 0.5
0.052 1.64E-05
0.251 8.03E-05
Fructose rates were measured for a number of conditions. The constant pH data are plotted as
under.
Figure 21: ln Rate vs ln Cf
y = 1.0064x - 8.0406
-16
-15
-14
-13
-12
-11
-10
-9
-8
-3.2-3-2.8-2.6-2.4-2.2-2-1.8-1.6-1.4-1.2
Ln R
ate
Ln Cf
pH 2 103.6 degC
ph 0.5 104.1 degC
50
Figure 22: ln Rate vs ln H+ (Fructose)
Figure 21 and Figure 22 establish that fructose dehydration rates are first order in both protons
and fructose. Since orders have been established, we may proceed to calculate rate constants and
consequently barriers and other kinetic constants for fructose dehydration.
TABLE 10: SECOND ORDER RATE CONSTANTS FOR FRUCTOSE CALCULATED FROM EXPERIMENTAL DATA.
Temperature ℃ Rate constant [l/mol.min]
103.6 0.0006
112.5 0.002
122.4 0.006
104.1 0.001
The Arrhenius plot better represents temperature effects on rate. The Arrhenius plot for fructose
shown in Figure 23.
y = 1.0962x - 8.1352
-16
-15
-14
-13
-12
-11
-10
-9
-8
-5.67-5.17-4.67-4.17-3.67-3.17-2.67-2.17-1.67-1.17-0.67Ln
Rat
e
Ln H+
104 degC, 1% fructose
104 degC, 5% fructose
51
Figure 23: Arrhenius plot (Fructose)
The activation energy obtained from this data is 140.45 ± 16.306 kJ/mol. Literature data for
fructose dehydrations with sulfuric acid indicate an Ea of approximately 136 kJ/mol.
-8
-7.5
-7
-6.5
-6
-5.5
-5
-4.5
0.0025 0.00255 0.0026 0.00265 0.0027
Ln k
1/T
52
5.5.3 Xylose
Figure 24: ln Rate vs Ln Cx
Xylose data were collected for only one pH value and two temperatures. The rate increase
facilitated by an increase in temperature from 123 to 132 ℃ is clear in Figure 24. Both sets of
data were obtained for a pH of 2. Xylose dehydration is also found to be first order in sugar and
protons. Xylose rates used in the analysis are shown in Table 9.
y = 1.0177x - 11.096
-14
-13.5
-13
-12.5
-12
-11.5
-11
-10.5
-10
-2.9-2.7-2.5-2.3-2.1-1.9-1.7-1.5-1.3-1.1-0.9
Ln R
ate
Ln Cx
pH 2 123 degC pH 2 132 degC
53
TABLE 11: RATES OF FURFURAL FORMATION FROM XYLOSE FEED SOLUTION.
Temperature ℃ Xylose (wt %) pH Rate (mol/l.min)
123 1 2 9.577E-07
5 2 4.863E-06
132 1 2 3.632E-06
5 2 1.444E-05
TABLE 12: SECOND ORDER RATE CONSTANTS FOR XYLOSE CALCULATED FROM EXPERIMENTAL RATE DATA
Temperature ℃ Rate constant [l/mol.min]
123 2.48E-05
132 6.87E-05
The Arrhenius plot for xylose is shown in Figure 24.
Figure 25: Arrhenius Plot (Xylose).
y = -19737x + 43.862
-7
-6.5
-6
-5.5
-5
-4.5
0.00244 0.00246 0.00248 0.0025 0.00252 0.00254 0.00256
Ln k
1/T
54
Due to limited experimental data, confidence intervals cannot be calculated for xylose. Ea was
found to be 164.092 J/mol and A was found to be 1.119E+19 ml/(mol.min).
TABLE 13: 2ND ORDER RATE CONSTANTS FOR SUGAR DEHYDRATION AT FIXED TEMPERATURES.
Temperature℃ Rate Constant k [l/(mol.min)]
Glucose Fructose Xylose
123 4.581E-05 0.007 0.002
132 0.0001 0.018 0.007
144 0.0003 0.045 0.022
We can compare rates of sugar dehydration over the temperature range being studied by
observing rate constants. Rate information collected over different temperatures were used to
obtain rate constants. From rate constants for each sugar, we are able to generate Arrhenius plots.
Since rates need to be compared at a fixed temperature to be of any significance to the
comparison, we use the Arrhenius plots to calculate values of k for that temperature. Although
limited data were collected at temperatures 123,132 and 141℃ for all sugars, it was possible to
calculate rate constants at these temperatures graphically and present them in Table 10. From this
table it is clear that lowest rates were seen for glucose dehydration. Fructose dehydration rates
were significantly higher.
55
Figure 26: Comparison of sugar dehydration rates based on rate constant. Rate constants taken from
Table 10.
Rates of dehydration are found to be greatest for fructose. The large relative rate difference
between fructose and glucose have already been explained in previous sections. Xylose, much
like fructose and glucose undergo several decomposition reactions at higher temperatures.
Dehydration of pentose sugars (xylose) to furfural is different from the mechanisms described
for dehydration of hexose sugars (fructose and glucose) to HMF. The intermediate 2, 5-
anhydride is formed through the cyclic xylofuranose molecule. Comparison of kinetic parameters
4.58E-05 0.0001 0.0003
0.007
0.018
0.045
0.002
0.007
0.022
0.00E+00
5.00E-03
1.00E-02
1.50E-02
2.00E-02
2.50E-02
3.00E-02
3.50E-02
4.00E-02
4.50E-02
5.00E-02
123℃ 132℃ 144℃
RA
TE C
ON
STA
NT
K [
L/M
OL.
MIN
]
Glucose Fructose Xylose
56
of xylose to the other sugars is difficult owing to the larger confidence intervals in data. Rates of
dehydration of xylose are, however greater than glucose as seen in Figure 26. Literature studies
that detail xylose dehydration show that the steps leading to anhydride (intermediate) and
consequently to furans through the cyclic route are faster than rates seen with open-chain xylose
since a very large fraction of open-chains would have to be formed for the reaction at
experimental temperatures. Glucose dehydration rates are lowest and rely on isomerization to
fructose before forming HMF. Rates as observed (fructose > xylose > glucose) were explained
and confirmed in the literature.
57
Chapter VI
Conclusion and Perspectives
The goal of this project was to develop a method and use it to generate reliable kinetic
data for use in industrial applications in the processing of HMF and furfural that have potential
as platform chemicals. Reliable data were generated using the reactor built through the
systematic heuristics based approach. Special considerations were made to account for changing
variables, such as density and its effect on rates. Catalyst loading was quantified based on proton
concentration (dissociation dependent on temperature and solvent) and the difference in proton
concentration was evaluated. The first part of the work involved designing a system based on
literature, assigning convenient and required process/mixing parameters and collecting
experimental data. Experimental data were collected between 123 and 141℃ . Despite low
conversions, rate data were measured for all three sugars. The second part of the study compared
the sugars based on the measured and calculated kinetic constants. Higher rates of fructose
dehydration as reported in other literature studies were confirmed through the analysis.
Assisting the renewables sector is an important objective of catalysis research. This study
explored the use of the homogeneous catalyst sulfuric acid in the dehydration reaction. Similar
setups may be used to study alternative catalysts (homogeneous or heterogeneous) and reactions
(multiphase). Only research into different catalysts can reveal novel catalysts that are selective,
inexpensive to operate and efficient at larger yields.
58
References
1. Klass, Donald L. Biomass as a nonfossil fuel source. No. CONF-790415-P8. American
Chemical Society, Washington, DC, 1981.
2. Klass, Donald L. "Biomass for renewable energy and fuels. Encyclopedia of
Energy." Elsevier Inc 1 (2004): 193-211.
3. Klass, Donald L. Biomass for renewable energy, fuels, and chemicals. Academic press,
1998.
4. Klass, D. L., and G. H. Emert. "‘Fuels from Biomass and Wastes. Ann Arbor Science."
(1981).
5. Wyman, Charles E., et al. "Hydrolysis of cellulose and hemicellulose."Polysaccharides:
Structural diversity and functional versatility 1 (2005): 1023-1062.
6. Wyman, Charles E., and Norman D. Hinman. "Ethanol." Applied Biochemistry and
Biotechnology 24.1 (1990): 735-753.
7. Wyman, Charles E. "Ethanol from lignocellulosic biomass: technology, economics, and
opportunities." Bioresource Technology 50.1 (1994): 3-15.
8. Tyson, Karin Shaine. Fuel cycle evaluations of biomass-ethanol and reformulated
gasoline. Volume 1. No. NREL/TP--463-4950. National Renewable Energy Lab., Golden,
CO (United States); Oak Ridge National Lab., TN (United States); Pacific Northwest
Lab., Richland, WA (United States), 1993.
9. Lynd, Lee R., et al. "Fuel ethanol from cellulosic
biomass."Science(Washington) 251.4999 (1991): 1318-1323.
10. Huber, George W., Sara Iborra, and Avelino Corma. "Synthesis of transportation fuels
from biomass: chemistry, catalysts, and engineering."Chemical reviews 106.9 (2006):
4044-4098.
11. Lange, Jean‐Paul, et al. "Furfural—a promising platform for lignocellulosic
biofuels." ChemSusChem 5.1 (2012): 150-166.
12. Wang, Tianfu, Michael W. Nolte, and Brent H. Shanks. "Catalytic dehydration of C 6
carbohydrates for the production of hydroxymethylfurfural (HMF) as a versatile platform
chemical." Green Chemistry 16.2 (2014): 548-572.
13. Phillips, J. R., et al. "Biological production of ethanol from coal synthesis gas."Applied
biochemistry and biotechnology 39.1 (1993): 559-571.
14. Balat, M. "Mechanisms of thermochemical biomass conversion processes. Part 3:
reactions of liquefaction." Energy Sources, Part A 30.7 (2008): 649-659.
15. Replace
16. Huber, George W., et al. "Production of liquid alkanes by aqueous-phase processing of
biomass-derived carbohydrates." Science 308.5727 (2005): 1446-1450.
17. Manzer, Leo E. "Biomass derivatives: a sustainable source of chemicals." ACS
symposium series. Vol. 921. Oxford University Press, 2006.
18. Perlack, Robert D., et al. Biomass as feedstock for a bioenergy and bioproducts industry:
the technical feasibility of a billion-ton annual supply. Oak Ridge National Lab TN,
2005.
59
19. Zakrzewska, Małgorzata E., Ewa Bogel-Łukasik, and Rafał Bogel-Łukasik. "Ionic liquid-
mediated formation of 5-hydroxymethylfurfural A promising biomass-derived building
block." Chemical reviews 111.2 (2010): 397-417.
20. Mosier, Nathan, et al. "Features of promising technologies for pretreatment of
lignocellulosic biomass." Bioresource technology 96.6 (2005): 673-686.
21. Kobayashi, Hirokazu, and Atsushi Fukuoka. "Synthesis and utilisation of sugar
compounds derived from lignocellulosic biomass." Green Chemistry 15.7 (2013): 1740-
1763.
22. Zakzeski, Joseph, et al. "The catalytic valorization of lignin for the production of
renewable chemicals." Chemical reviews 110.6 (2010): 3552-3599.
23. Corma, Avelino, Sara Iborra, and Alexandra Velty. "Chemical routes for the
transformation of biomass into chemicals." Chemical Reviews 107.6 (2007): 2411-2502.
24. Roberts, Virginia, et al. "Towards quantitative catalytic lignin
depolymerization."Chemistry-A European Journal 17.21 (2011): 5939-5948.
25. Zhao, Chen, et al. "Aqueous-phase hydrodeoxygenation of bio-derived phenols to
cycloalkanes." Journal of Catalysis 280.1 (2011): 8-16.
26. Azadi, Pooya, et al. "Catalytic conversion of biomass using solvents derived from
lignin." Green Chemistry 14.6 (2012): 1573-1576.
27. Moreau, Claude, et al. "Selective preparation of furfural from xylose over microporous
solid acid catalysts." Industrial Crops and Products 7.2 (1998): 95-99.
28. Moreau, Claude, et al. "Dehydration of fructose to 5-hydroxymethylfurfural over H-
mordenites." Applied Catalysis A: General 145.1 (1996): 211-224.
29. Plow, R. H., et al. "Rotary Digester in Wood Saccharification." Industrial & Engineering
Chemistry 37.1 (1945): 36-43.
30. Lourvanij, Khavinet, and Gregory L. Rorrer. "Reactions of aqueous glucose solutions
over solid-acid Y-zeolite catalyst at 110-160. degree. C." Industrial & engineering
chemistry research 32.1 (1993): 11-19.
31. Huber, George W., et al. "Production of liquid alkanes by aqueous-phase processing of
biomass-derived carbohydrates." Science 308.5727 (2005): 1446-1450.
32. Hu, Lei, et al. "Catalytic conversion of biomass-derived carbohydrates into fuels and
chemicals via furanic aldehydes." RSC Advances 2.30 (2012): 11184-11206.
33. Qian, Xianghong. "Mechanisms and Energetics for Brønsted Acid-Catalyzed Glucose
Condensation, Dehydration and Isomerization Reactions." Topics in Catalysis 55.3-4
(2012): 218-226.
34. Karinen, Reetta, Kati Vilonen, and Marita Niemelä. "Biorefining: heterogeneously
catalyzed reactions of carbohydrates for the production of furfural and
hydroxymethylfurfural." ChemSusChem 4.8 (2011): 1002-1016.
35. Ståhlberg, Tim, et al. "Synthesis of 5‐(Hydroxymethyl) furfural in Ionic Liquids: Paving
the Way to Renewable Chemicals." ChemSusChem 4.4 (2011): 451-458.
36. Ranoux, Adeline, et al. "5-Hydroxymethylfurfural synthesis from hexoses is
autocatalytic." ACS Catalysis 3.4 (2013): 760-763.
37. Kuster, B. F. M. "5‐Hydroxymethylfurfural (HMF). A review focussing on its
manufacture." Starch‐Stärke 42.8 (1990): 314-321.
38. Alamillo, Ricardo, et al. "The selective hydrogenation of biomass-derived 5-
hydroxymethylfurfural using heterogeneous catalysts." Green Chemistry 14.5 (2012):
1413-1419.
60
39. Tong, Xinli, Yang Ma, and Yongdan Li. "Biomass into chemicals: conversion of sugars
to furan derivatives by catalytic processes." Applied Catalysis A: General 385.1 (2010):
1-13.
40. Román-Leshkov, Yuriy, et al. "Production of dimethylfuran for liquid fuels from
biomass-derived carbohydrates." Nature 447.7147 (2007): 982-985.
41. Balakrishnan, Madhesan, Eric R. Sacia, and Alexis T. Bell. "Etherification and reductive
etherification of 5-(hydroxymethyl) furfural: 5-(alkoxymethyl) furfurals and 2, 5-bis
(alkoxymethyl) furans as potential bio-diesel candidates." Green Chemistry 14.6 (2012):
1626-1634.
42. Kraus, George A., and Tezcan Guney. "A direct synthesis of 5-alkoxymethylfurfural
ethers from fructose via sulfonic acid-functionalized ionic liquids." Green Chemistry 14.6
(2012): 1593-1596.
43. Lew, Christopher M., Nafiseh Rajabbeigi, and Michael Tsapatsis. "One-pot synthesis of
5-(ethoxymethyl) furfural from glucose using Sn-BEA and Amberlyst
catalysts." Industrial & Engineering Chemistry Research 51.14 (2012): 5364-5366.
44. Wang, Tianfu, Michael W. Nolte, and Brent H. Shanks. "Catalytic dehydration of C 6
carbohydrates for the production of hydroxymethylfurfural (HMF) as a versatile platform
chemical." Green Chemistry 16.2 (2014): 548-572.
45. Hu, Wen-Jing, et al. "Repression of lignin biosynthesis promotes cellulose accumulation
and growth in transgenic trees." Nature biotechnology 17.8 (1999): 808-812.
46. O'SULLIVAN, ANTOINETTE C. "Cellulose: the structure slowly
unravels."Cellulose 4.3 (1997): 173-207.
47. Hsu, T. A.; Ladisch, M. R.; Tsao, G. T. Chem. Technol. 1980, 10, 315.
48. Kuster, B. F. M. "5‐Hydroxymethylfurfural (HMF). A review focussing on its
manufacture." Starch‐Stärke 42.8 (1990): 314-321.
49. Mosier, Nathan S., Christine M. Ladisch, and Michael R. Ladisch. "Characterization of
acid catalytic domains for cellulose hydrolysis and glucose degradation." Biotechnology
and bioengineering 79.6 (2002): 610-618.
50. Kuster, B. F. M. "5‐Hydroxymethylfurfural (HMF). A review focussing on its
manufacture." Starch‐Stärke 42.8 (1990): 314-321.
51. Alonso, David Martin, Jesse Q. Bond, and James A. Dumesic. "Catalytic conversion of
biomass to biofuels." Green Chemistry 12.9 (2010): 1493-1513.
52. Cinlar, Basak, Tianfu Wang, and Brent H. Shanks. "Kinetics of monosaccharide
conversion in the presence of homogeneous Bronsted acids." Applied Catalysis A:
General 450 (2013): 237-242.
53. Bandura, Andrei V., and Serguei N. Lvov. "The ionization constant of water over wide
ranges of temperature and density." Journal of physical and chemical reference data 35.1
(2006): 15-30.
54. Marshall, William L., and Ernest V. Jones. "Second Dissociation Constant of Sulfuric
Acid from 25 to 350° Evaluated from Solubilities of Calcium Sulfate in Sulfuric Acid
Solutions1, 2." The Journal of Physical Chemistry 70.12 (1966): 4028-4040.
55. Aristidou, Aristos, and Merja Penttilä. "Metabolic engineering applications to renewable
resource utilization." Current Opinion in Biotechnology 11.2 (2000): 187-198.
56. Sonderegger, Marco, et al. "Fermentation performance of engineered and evolved xylose‐fermenting Saccharomyces cerevisiae strains." Biotechnology and bioengineering 87.1
(2004): 90-98.
61
57. Nandi, R., and S. Sengupta. "Microbial production of hydrogen: an overview."Critical
reviews in microbiology 24.1 (1998): 61-84.
58. Wyman, Charles E., et al. "Comparative sugar recovery data from laboratory scale
application of leading pretreatment technologies to corn stover."Bioresource
technology 96.18 (2005): 2026-2032.
59. Jambusaria, R. B., and S. P. Potnis. "Bisphenol—Furfural. A High-Temperature
Thermosetting Resin." Polymer Journal 6.5 (1974): 333-340.
60. Sitthisa, Surapas, Wei An, and Daniel E. Resasco. "Selective conversion of furfural to
methylfuran over silica-supported Ni Fe bimetallic catalysts." Journal of Catalysis 284.1
(2011): 90-101.
61. Carpita, Nicholas C., and David M. Gibeaut. "Structural models of primary cell walls in
flowering plants: consistency of molecular structure with the physical properties of the
walls during growth." The Plant Journal 3.1 (1993): 1-30.
62. Román-Leshkov, Yuriy, Juben N. Chheda, and James A. Dumesic. "Phase modifiers
promote efficient production of hydroxymethylfurfural from fructose."Science 312.5782
(2006): 1933-1937.
63. Klingler, D., and H. Vogel. "Influence of process parameters on the hydrothermal
decomposition and oxidation of glucose in sub-and supercritical water." The Journal of
Supercritical Fluids 55.1 (2010): 259-270.
64. Asghari, Feridoun Salak, and Hiroyuki Yoshida. "Dehydration of fructose to 5-
hydroxymethylfurfural in sub-critical water over heterogeneous zirconium phosphate
catalysts." Carbohydrate research 341.14 (2006): 2379-2387.
65. Mosier, Nathan S., et al. "Characterization of dicarboxylic acids for cellulose
hydrolysis." Biotechnology progress 17.3 (2001): 474-480.
66. Heeres, H. J., L. P. B. M. Janssen, and B. Girisuta. "A Kinetic Study on the Conversion
of Glucose to Levulinic Acid." (2006).
67. Van Dam, H. E., A. P. G. Kieboom, and H. Van Bekkum. "The conversion of fructose
and glucose in acidic media: formation of hydroxymethylfurfural."Starch‐Stärke 38.3
(1986): 95-101.
68. Antal, Michael Jerry, William SL Mok, and Geoffrey N. Richards. "Mechanism of
formation of 5-(hydroxymethyl)-2-furaldehyde from D-fructose and
sucrose."Carbohydrate research 199.1 (1990): 91-109.
62
Vita